The human breast cancer resistance protein (BCRP/MXR/ABCG2) is a relatively new ATP-binding cassette transporter originally cloned from drug-selected human cancer cell lines and human placenta (Allikmets et al., 1998; Doyle et al., 1998; Miyake et al., 1999). Like P-glycoprotein (P-gp), BCRP confers high levels of resistance to anthracyclines, mitoxantrone, and the camptothecins by enhancing drug efflux from the cell (Litman et al., 2000; Bates et al., 2001; Ejendal and Hrycyna, 2002). Indeed, in acute leukemia, BCRP may play an important role in resistance to flavopiridol (Robey et al., 2001b; Nakanishi et al., 2003b). In addition to its role in resistance to chemotherapeutic agents, BCRP actively transports structurally diverse organic drugs, conjugated or unconjugated, such as estrone-3-sulfate, 17β-estradiol 17-(β-d-glucuronide), and methotrexate (Volk et al., 2002; Chen et al., 2003; Imai et al., 2003; Suzuki et al., 2003; Volk and Schneider, 2003). Sequence analysis of BCRP cDNA revealed mutations at position 482 in several drug-selected cell lines (Honjo et al., 2001; Allen et al., 2002). Subsequent studies showed that position 482 is important in determining substrate specificity of BCRP (Honjo et al., 2001; Allen et al., 2002; Volk et al., 2002; Chen et al., 2003; Robey et al., 2003). For instance, wild-type BCRP (482R) does not efflux rhodamine 123 but the mutants 482T and 482G can readily transport rhodamine (Honjo et al., 2001). BCRP is prominently expressed in placental syncytiotrophoblasts, in the epithelium of the small intestine, and in the liver canalicular membrane (Maliepaard et al., 2001; Doyle and Ross, 2003). In fact, BCRP transcript is expressed in the intestine in greater amounts than P-gp (Taipalensuu et al., 2001). This strategic and substantial tissue localization implies that BCRP also functions as a protective drug efflux pump in the placenta and the intestine. Indeed, Jonker et al. (2000) have shown that treatment with the BCRP inhibitor GF120918 (also a P-gp inhibitor), decreases plasma clearance and hepatobiliary excretion of the anticancer agent topotecan and increases absorption of this drug from the small intestine in P-gp knockout mice. In pregnant GF120918-treated P-gp-deficient mice, the relative fetal concentration of topotecan was 2-fold higher than that in pregnant vehicle-treated mice. These observations suggest that Bcrp1 (the murine ortholog of BCRP) mediates apically directed drug transport, reduces drug bioavailability, and protects fetuses against therapeutic agents. A recent clinical study showed that inhibition of BCRP significantly increases the oral bioavailability of topotecan in cancer patients from 40 to 97% (Kruijtzer et al., 2002). Thus, BCRP is important in determining absorption, distribution, and elimination of drugs that are substrates for this transporter.

Many of the drugs transported by BCRP (listed above) are also substrates of P-gp. Some have suggested that the substrate selectivity of BCRP substantially overlaps with that of P-gp (Litman et al., 2000; Doyle and Ross, 2003). Therefore, to comprehensively evaluate the importance of BCRP in the in vivo disposition of drugs, it is important to determine the substrate profile of BCRP. Hence, we have asked whether the HIV protease inhibitors (HPIs), excellent substrates of P-gp, are substrates of BCRP. In this article, we present data that addresses this question.

Materials and Methods

Materials. Ritonavir was a gift from National Institutes of Health AIDS Research and Reference Reagent Program (National Institutes of Health, Bethesda, MD). Saquinavir and [¹⁴C]saquinavir (27.71 mCi/mmol) were from Roche Diagnostics (Nutley, NJ) and amprenavir was from GlaxoSmithKline (Research Triangle Park, NC). Nelfinavir and [³H]nelfinavir (60 mCi/mmol) were from Pfizer Global Research and Development (San Diego, CA). [³H]Ritonavir (1.2 Ci/mmol), [³H]amprenavir (2.0 Ci/mmol), and [³H]mitoxantrone (1.5 Ci/mmol) were purchased from Moravek Biochemicals (Brea, CA). Indinavir was a gift from Merck (West Point, PA). Mitoxantrone (MX) hydrochloride was from Sigma-Aldrich (St. Louis, MO). Fumitremorgin C (FTC) was a generous gift from Dr. Susan E. Bates (National Cancer Institutes, Bethesda, MD). High-performance liquid chromatography grade DMSO was obtained from Fisher Scientific Co. (Pittsburgh, PA) and used as the solvent for making stock solutions of all the drugs and FTC. Eagle's minimal essential medium, penicillin-streptomycin-glutamine solution, and trypsin-EDTA were purchased from American Type Culture Collection (Manassas, VA). Dulbecco's modified Eagle's phenol-free low-glucose medium (DMEM), phosphate-buffered saline (PBS), trypsin-EDTA solution, and fetal bovine serum (FBS) were obtained from Invitrogen (Carlsbad, CA).

Cell Culture and Whole Cell Lysate Preparation. HEK293 cells stably transfected with pcDNA empty vector and cDNAs coding for wild-type BCRP (482R) and its two mutants (482T and 482G) were obtained from Dr. Susan E. Bates (National Cancer Institute) (Robey et al., 2003). All the cell lines were grown and maintained in minimal essential medium supplemented with 10% FBS, 2 mM l-glutamine, 100 units/ml penicillin, 100 μg/ml streptomycin, and 0.5 mg/ml G418 (Mediatech, Herndon, VA) at 37°C in a 5% CO₂ incubator. Cells were grown to 80 to 90% confluence and treated with trypsin-EDTA before harvesting for subculturing or efflux assays. Only cells within six passages were used in subsequent transport experiments.

Whole cell lysates were prepared as follows. Briefly, cell pellet from 5 × 10⁶ cells was suspended in approximately 100 μl of lysis buffer containing 50 mM Tris/HCl, pH 7.5, 10 mM MgCl₂, 50 μg/ml phenylmethylsulfonyl fluoride, 0.25% (w/v) SDS, 250 μg/ml DNase I (Invitrogen), and protease inhibitors (Complete; Roche Diagnostics). The mixture was then incubated on ice for 1 h with gentle vortexing occasionally and centrifuged at 2400g for 10 min at 4°C. The supernatant was subjected to immunoblotting as described below. Protein concentrations were determined using a modified Lowry assay (Peterson, 1979) and bovine serum albumin as standard.

SDS-Polyacrylamide Gel Electrophoresis and Immunoblotting. Gel electrophoresis of whole cell lysates from various cell lines was carried out using 10% SDS-polyacrylamide mini gels in a Mini-protein II electrophoresis cell (Bio-Rad, Hercules, CA). Standard Western blot procedure was followed. Briefly, proteins were transferred on Immuno-P nitrocellulose membranes (Millipore Corporation, Billerica, MA) using 25 mM Tris, 192 mM glycine, and 20% methanol buffer. The blot was blocked with 5% skim milk in TBS-T buffer (10 mM Tris/HCl, pH 7.5, 0.15 mM NaCl, and 0.05% Tween 20) for 1 h at room temperature. The blot was incubated with mAb BXP-21, which recognizes an internal epitope in the nucleotide-binding domain (amino acids 271–396) of BCRP (Maliepaard et al., 2001) (Kamiya Biomedical, Seattle, WA), at 1:500 dilution as the primary antibody for 2 h at room temperature followed by washing the blot three times with TBS-T. The blot was then incubated with the secondary antibody, goat anti-mouse-horseradish peroxidase conjugate (dilution 1:10,000) (Bio-Rad) for 1 h at room temperature. The blot was then washed three times with TBS-T buffer and developed using a Supersignal West Pico chemiluminescence detection kit (Pierce Chemical, Rockford, IL). Relative levels of BCRP expression were quantitated by densitometric analysis of the immunoblot with Chemi Doc system (Bio-Rad) and Quantity One software version 4.3.0. For immunoblotting of P-gp, MRP1, and MRP2, the same procedure as described above was used except that the primary antibody used was mAb F4 (Kamiya Biomedical) for P-gp (dilution 1:1,000), mAb QCRL-1 (Signet Laboratories, Dedham, MA) for MRP1 (dilution 1:500) and mAb M₂III-6 (Alexis Biochemicals, San Diego, CA) for MRP2 (dilution 1:100).

Flow Cytometric Efflux Assays. The efflux assays were essentially the same as described previously (Wang et al., 2000; Robey et al., 2001a) with minor modifications. Briefly, BCRP-expressing or vector control HEK cells were harvested and suspended in incubation buffer (DMEM supplemented with 5% FBS and 5 mM HEPES buffer) at cell concentration of approximately 10⁶ cells per reaction in a 1-ml volume. In the accumulation phase, cell suspensions were coincubated with 10 μM MX and variable concentrations (0–150 μM) of HPIs or 10 μM FTC as a positive control for 30 min at 37°C to allow accumulation of drugs. Cells were then immediately transferred on ice, washed once with ice-cold PBS, and resuspended in 1 ml of incubation buffer containing only the respective HPIs or FTC at their respective concentrations as in the accumulation phase and incubation was continued for 1 h at 37°C to allow maximum efflux of MX (efflux phase). Cells were washed once and finally resuspended in 1 ml of ice-cold PBS. Intracellular MX fluorescence was measured within 1 h with a 488-nm argon laser and a 650-nm bandpass filter in a BD FACSCAN flow cytometer. Cells were kept on ice until intracellular MX fluorescence was read. Ten thousand (10⁴) events were collected for all the samples. Cell debris was eliminated by gating on forward versus side scatter. Cells in medium containing vehicle [4% (v/v) DMSO] yielded the blank histogram, a measure of cellular autofluorescence. Cells in medium containing MX alone or medium containing MX and FTC or HPI generated the control and FTC or HPI histograms, respectively. The difference (ΔF) between the median fluorescence of the control histogram and the median fluorescence of the FTC or HPI histogram was used to express inhibition of MX efflux by FTC or HPIs, respectively. The maximum concentration of DMSO used as vehicle for HPIs was 4% (v/v) in all the efflux assays. No effects of the vehicle on MX efflux were observed at this concentration.

Inhibition of BCRP-Mediated Efflux of [³H]MX by HPIs. Direct efflux of [³H]MX in BCRP-expressing or vector control HEK cells in the presence and absence of cold HPIs was also measured to examine whether HPIs inhibit BCRP-mediated efflux of [³H]MX. Briefly, 0.5 × 10⁶ cells were incubated with [³H]MX (20 nM, 0.03 μCi) and various concentrations of HPIs (25 or 100 μM) in 0.5 ml of incubation buffer (DMEM supplemented with 5 mM HEPES buffer) for 30 min at 37°C. To avoid possible protein binding of [³H]MX, FBS was omitted from the incubation buffer. The cells were then transferred on ice, washed once with ice-cold PBS, and resuspended in 0.5 ml of incubation buffer in the presence of respective HPIs for 1 h at 37°C. Efflux from the cells was terminated by washing once in ice-cold PBS. The cell pellet was lysed with 1 ml of 1% SDS, and 900 μl of the lysate was subjected to counting in a scintillation counter. Counts were normalized to protein concentration that was measured using the remaining lysate by the modified Lowry assay. The intracellular [³H]MX was calculated based on radioactivity associated with the cell pellet and expressed as picomoles of MX per microgram of protein. The concentration of DMSO used as vehicle for HPIs was 0.5% (v/v). No effects of the vehicle on [³H]MX efflux were observed at this concentration.

Efflux of Radiolabeled HPIs. Direct efflux assays using radiolabeled HPIs were carried out to confirm whether the HPIs are substrates of BCRP as follows. Briefly, the BCRP-expressing or vector control HEK cells (0.5 × 10⁶ cells) were incubated with a radiolabeled HPI at a desired concentration (with or without FTC) in 0.5 ml of incubation buffer (DMEM supplemented with 5 mM HEPES buffer) for 30 min at 37°C to allow maximum accumulation of HPIs in cells. The concentrations (specific activities) of [³H]ritonavir, [¹⁴C]saquinavir, [³H]amprenavir, and [³H]nelfinavir used in the accumulation phase were 0.5 μM (0.6 μCi/ml), 1 μM (0.028 μCi/ml), 0.25 μM (0.5 μCi/ml), and 1 μM (0.06 μCi/ml), respectively. The cells were then transferred on ice, washed once with ice-cold PBS, and resuspended in 0.5 ml of incubation buffer (with or without FTC), and incubation was continued for 1 h at 37°C to allow maximum efflux of the drug. The efflux reactions were stopped by washing the cells once with ice-cold PBS. Cell pellet was dissolved in 1 ml of 1% SDS, and 900 μl of the lysate was subjected to counting in a scintillation counter. Counts were normalized to protein concentration that was measured by the modified Lowry assay using the remaining lysate. The intracellular levels of HPIs were calculated based on radioactivity associated with the cell pellet and expressed as picomoles of drug per microgram of protein or femtomoles of drug per microgram of protein. We tried various drug loading and efflux times and found that the experiments produced the optimal efflux activity with 30-min drug loading and 1 h efflux.

Data Analysis. IC₅₀ values representing the inhibitory effectiveness of HPIs on BCRP-mediated MX efflux in the flow cytometric efflux assays were calculated by fitting the following model to the data (Fig. 3) using nonlinear regression (WinNonLin software version 3.2 (Pharsight, Mountain View, CA): where ΔF and ΔF_max are inhibition and the maximal inhibition, respectively. IC₅₀ is the concentration of HPIs leading to half-maximal inhibition of BCRP activity, C is the concentration of HPI, and γ is the slope factor. To elucidate whether the IC₅₀ values of the HPIs (Table 1) for wild-type BCRP (482R) are statistically different from the IC₅₀ values for the BCRP mutants (482T and 482G), we compared the IC₅₀ values for 482R with the values for 482T or 482G, one pair at a time for each HPI, by Student's t test. A p value of <0.05 was considered significant.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Effects of HPIs (0–150 μM) on MX efflux in HEK cells. Graphs represent shift in intracellular MX fluorescence (ΔF) versus concentrations of various HPIs in the HEK cell expressing wild-type BCRP (482R) (A) and its two mutants 482T (B) and 482G (C). Ritonavir is represented by solid squares (▪), saquinavir by solid triangles (▴), nelfinavir by solid circles (•), indinavir by open circles (○), and amprenavir by open squares (□). The data points are mean ± S.D. of three to four independent experiments.

Results

Expression of BCRP in HEK Cells. Whole cell lysates prepared from various HEK cell lines stably expressing wild-type BCRP (482R) and its two mutants (482T and 482G) were subjected to immunoblotting for analysis of BCRP expression. Figure 1 shows substantial expression of BCRP in the HEK cells after 30-s exposure. 482R and 482G displayed comparable levels of expression. However, 482T was expressed in amounts approximately 20% lower than 482R and 482G. Similar expression pattern of BCRP in HEK cells has been reported previously (Robey et al., 2003). Because HPIs are known substrates for P-gp, MRP1, or MRP2, to rule out any possible contribution of these transporters, the same protein samples were subjected to immunoblotting for P-gp, MRP1, and MRP2. No P-gp, MRP1, or MRP2 was detected in the samples after a brief exposure for 1 min. Faint bands for these transporters occurred only upon prolonged exposure of the blot for approximately 2 h. The levels of expression of P-gp, MRP1, and MRP2 were comparable in both the BCRP-expressing and vector-control cells (data not shown). These bands represent the endogenous expression levels of these transporters. Thus, relative to the expression of BCRP, the HEK cells express little endogenous P-gp, MRP1, or MRP2.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

BCRP expression, as determined by immunoblotting, in whole cell lysates (20 μg) of HEK cells stably expressing wild-type BCRP(482R) and its two mutants (482T and 482G) or transfected with the control vector (pcDNA). The relative expression of BCRP was 1.0, 1.18, and 1.27 (arbitrary unit) for 482T, 482R, and 482G, respectively.

Effects of HPIs on BCRP-Mediated MX Efflux. MX is a fluorescent compound and a well known BCRP substrate. MX is not a substrate of MRP1 (Litman et al., 2000). Although MX is a substrate of both P-gp and BCRP (Litman et al., 2000), we have already shown that the HEK cells express little of endogenous P-gp. Hence, MX can be used as a model substrate for measuring BCRP transport activity in the HEK cells using flow cytometry. We observed that the level of intracellular MX was much greater in HEK cells transfected with pcDNA empty vector than in the BCRP-overexpressing cells (data not shown), and the reduction in intracellular MX in the BCRP expressing cells could be abrogated by addition of 10 μM FTC, a specific BCRP inhibitor (Fig. 2A). This suggests that the decrease in intracellular MX is mediated by BCRP. Therefore, the change in MX fluorescence (ΔF) upon addition of BCRP inhibitors can be used to express inhibition of BCRP activity as described under Materials and Methods. In preliminary studies, we found that 10 μM MX in the presence and absence of 10 μM FTC produced ΔF values that were optimal for determining BCRP activity (Fig. 2A). Hence, concentrations of 10 μM MX and 10 μM FTC were used in all the subsequent flow cytometric efflux assays.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Flow cytometric MX efflux assays. Efflux assays were performed with 30-min accumulation of MX and 1-h efflux as described under Materials and Methods. Representative histograms are shown for the HEK cells expressing BCRP (482R). The solid histograms represent autofluorescence of the cells. A, intracellular MX fluorescence in the absence (black bold line) or presence (gray line) of 10 μM FTC. B, intracellular MX fluorescence in the absence (black bold line) and presence of various concentrations (10–150 μM) of nelfinavir.

We first examined whether HPIs are BCRP inhibitors. The effect of various concentrations (0–150 μM) of five HPIs (ritonavir, saquinavir, nelfinavir, indinavir, and amprenavir) on intracellular MX fluorescence in HEK cell lines expressing wild-type BCRP and its two mutants was determined by flow cytometry. A typical inhibition profile for wild-type BCRP (482R) with nelfinavir is shown in Fig. 2B. Ritonavir, saquinavir, and nelfinavir significantly increased intracellular MX fluorescence in a concentration-dependent manner in the cells expressing wild-type BCRP and its two mutants (Fig. 3). In contrast, indinavir and amprenavir did not inhibit any isoforms of the BCRP protein, even at a high concentration of 150 μM. The same treatments of the vector control cell line with HPIs did not show any significant shift in the intracellular MX fluorescence (data not shown). These results demonstrate that ritonavir, saquinavir, and nelfinavir inhibit BCRP-mediated efflux of MX, whereas indinavir and amprenavir are not BCRP inhibitors.

For any given BCRP protein, IC₅₀ values of nelfinavir, ritonavir, and saquinavir were not significantly different (Table 1). However, the IC₅₀ values of all three drugs were significantly lower for the wild-type BCRP (482R) than the mutants (482T and 482G) but not significantly different within the two mutants. Frequently, a decrease in ΔF values at higher HPIs concentrations was observed (Fig. 3). This decrease was likely due to the cytotoxic effect of the HPIs on the cells exposed at higher concentrations. Interestingly, the reduction was more pronounced for 482R cells than cells expressing the two mutants. This is consistent with the observation that 482R seems to be more sensitive to inhibition by the HPIs than the mutants 482T and 482G.

Inhibition of [³H]MX Efflux by HPIs. To further confirm that ritonavir, saquinavir, and nelfinavir are effective inhibitors of BCRP, the effect of these drugs on the direct efflux of [³H]MX was measured in HEK cell lines transfected with wild-type BCRP cDNA and the control vector. As expected, the BCRP (482R)-expressing cell line showed a significant reduction in intracellular [³H]MX level compared with the vector control cell line (Fig. 4). This reduction was abrogated by the addition of 10 μM FTC. In addition, ritonavir, saquinavir, and nelfinavir, at concentrations of 25 and 100 μM, reversed the reduction in intracellular [³H]MX (Fig. 4) in BCRP-expressing cells, indicating that the three HPIs are indeed inhibitors of BCRP. In contrast, indinavir and amprenavir, even at the high concentration of 100 μM, did not have a significant effect on [³H]MX efflux in the 482R cells, confirming that they are not inhibitors of BCRP.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Effects of HPIs on efflux of [³H]MX (20 nM) in the BCRP (482R) and vector control cells. The assays were carried out with 30-min accumulation of [³H]MX and 1-h efflux as described under Materials and Methods. Columns represent intracellular levels of [³H]MX as percentage of intracellular [³H]MX in the vector control cells. Intracellular [³H]MX levels were measured in the presence of various HPIs at concentrations of 25 μM (slashed columns) and 100 μM (solid columns). The checked column represents intracellular [³H]MX in 482R cells in the presence of 10 μM FTC, and the open column represents intracellular [³H]MX in 482R cells incubated with vehicle only (DMSO). Intracellular [³H]MX in the vector control cells incubated with MX and the vehicle was set as 100% (dotted column). The data are mean ± S.D. of three independent experiments. Statistical significance of the differences between the levels of intracellular MX in 482R cells with vehicle alone, and the rest of the data were determined by Student's t test (*, p < 0.05).

Efflux of Radiolabeled HPIs. To confirm whether the HPIs used in the above-mentioned studies are substrates for BCRP, direct efflux of these drugs in BCRP (482R)-expressing and vector control cells was measured using radiolabeled HPIs. Indinavir was not studied because it was not available to us in a radiolabeled form. The efflux of the four radiolabeled HPIs was not significantly different in the 482R and the vector control cells, indicating that all four HPIs are not substrates of BCRP (Fig. 5). Interestingly, intracellular [³H]ritonavir (Fig. 5A) and [¹⁴C]saquinavir levels (Fig. 5C) were increased approximately 40 and 200%, respectively, by addition of 10 μM FTC in both the vector control and 482R cells (solid columns). However, addition of FTC did not have significant influence on intracellular levels of [³H]amprenavir (Fig. 5B) and [³H]nelfinavir (Fig. 5D) in both the control and BCRP cells (solid columns). These results suggest that there may be an endogenous efflux transporter in the HEK cells for ritonavir and saquinavir, which can be inhibited by FTC.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Efflux of radiolabeled HPIs in the vector control and 482R cells incubated with 0.5 μM [³H]ritonavir (A), 0.25 μM [³H]amprenavir (B), 1 μM [¹⁴C]saquinavir (C), and 1 μM (D) [³H]nelfinavir. The detailed assay conditions with 30-min accumulation of radiolabeled HPIs and 1-h efflux are described under Materials and Methods. The open columns represent intracellular levels of HPIs alone. The solid columns represent intracellular levels of HPIs in the presence of 10 μM FTC. The data are as mean ± S.D. of three to five independent experiments. Statistical significance was determined by Student's t test (*, p < 0.05 and **, p < 0.01).

Discussion

HPIs, including saquinavir, indinavir, and ritonavir, are high-affinity substrates for P-gp, which limits their oral absorption and entry into the central nervous system (Kim et al., 1998a,b; Lee et al., 1998; Polli et al., 1999). Recent studies (Huisman et al., 2002; Williams et al., 2002) have shown that MRP2 also effectively transports saquinavir, ritonavir, and indinavir. Thus, low and/or variable oral bioavailability of HPIs could be explained in part by the high level expression of P-gp and MRP2 in the small intestine and the liver, where these transporters reduce absorption and increase intestinal and hepatobiliary clearance of drugs. BCRP shares many similarities with P-gp and MRP2 with respect to substrate specificity and tissue localization. Like P-gp and MRP2, BCRP is highly expressed in the luminal membranes in the small intestine and liver (Maliepaard et al., 2001; Doyle and Ross, 2003). In fact, BCRP transcript has been reported to be expressed in greater amount than P-gp in the intestine (Taipalensuu et al., 2001). Moreover, BCRP is also highly expressed in sites of clinical importance for HIV drug action such as the blood-brain barrier and placental barrier (Maliepaard et al., 2001; Cooray et al., 2002). Thus, BCRP would be expected to alter pharmacokinetic properties of HPIs if these drugs are BCRP substrates. This prompted us to investigate whether the HPIs are substrates and/or inhibitors of BCRP. The results of our study provide the first direct evidence that ritonavir, saquinavir, and nelfinavir are effective inhibitors of BCRP but indinavir and amprenavir are not and that none of the HPIs is a substrate for BCRP.

Two different analyses suggest that ritonavir, saquinavir, and nelfinavir inhibit BCRP. Ritonavir, saquinavir, and nelfinavir effectively increased intracellular MX fluorescence in the BCRP-expressing cells in a concentration-dependent manner as would be expected for BCRP inhibitors (Figs. 2 and 3). Indinavir and amprenavir did not significantly change intracellular MX fluorescence. Additional evidence that ritonavir, saquinavir and nelfinavir inhibit BCRP activity was obtained in the [³H]MX efflux assays. Ritonavir, saquinavir and nelfinavir effectively abrogated reduction of intracellular [³H]MX in the wild-type BCRP-overexpressing cells, whereas indinavir and amprenavir did not (Fig. 4). Given the fact that the five HPIs tested in this study have extremely diverse chemical structures and molecular sizes (their molecular weights range from 506 to 767), it remains to be investigated why some of the HPIs interact with BCRP but others do not and what determines the binding affinity of these compounds to the transporter.

We also found that the wild-type BCRP (482R) was significantly more sensitive than its mutants 482T and 482G to inhibition by the HPIs, suggesting that position 482 in BCRP may be a critical determinant for binding of the HPIs to BCRP (Table 1). Position 482 has been shown to be critical in determining substrate specificity of BCRP (Honjo et al., 2001; Allen et al., 2002; Volk et al., 2002; Chen et al., 2003; Robey et al., 2003). Thus, our data again support the evidence that mutations at position 482 in BCRP have a significant effect on ligand recognition by the transporter.

To determine whether HPIs are substrates of BCRP, we measured their efflux in cells expressing the wild-type BCRP. The efflux of radiolabeled ritonavir, saquinavir, nelfinavir, and amprenavir was not significantly different in the vector control and BCRP-overexpressing cells, suggesting that these drugs are not transported by BCRP (Fig. 5). These data are consistent with the results previously reported by other laboratories. Wang et al. (2003) analyzed cytotoxicity of nelfinavir in the drug-resistant MT-4/DOX₅₀₀ (BCRP-overexpressing cells) and the parental MT-4 cells and reported that cytotoxicity of nelfinavir was not reduced in MT-4/DOX₅₀₀ cells, a result to be expected if nelfinavir is not a substrate of BCRP. However, they did not directly measure transport of HPIs by the BCRP-overexpressing cells. Huisman et al. (2002) also found that Bcrp1, the murine homolog of BCRP, does not transport ritonavir, saquinavir, or indinavir. Together, our results and the data published by others indicate that the HPIs tested in this study are not substrates of BCRP. The data presented here are the first report where the inhibitory characteristics of HPIs to BCRP have been characterized in detail. Moreover, this is the first time that transport of HPIs by human BCRP has been measured directly. We noticed that the efflux of radiolabeled ritonavir and saquinavir by both the vector control HEK cells and the 482R cells could be inhibited by incubation with FTC, suggesting that there is an endogenous efflux transporter for ritonavir and saquinavir in the HEK cells that can be inhibited by FTC. Several HPIs investigated in this study are substrates of P-gp, MRP1, or MRP2 (Lee et al., 1998; Williams et al., 2002); however, it is unlikely that any of these pumps is the endogenous transporter for ritonavir and saquinavir in the HEK cells, based on the following evidence: first, we have shown that the HEKs cells express little endogenous P-gp, MRP1, or MRP2; and second, there is no evidence thus far to suggest that FTC is able to inhibit these transporters. Recently, Imai et al. (2003) also reported the existence of an endogenous transporter for MX in LLC-PK1 cells that can be inhibited by FTC. Whether the endogenous transporter found by Imai et al. (2003) and the one reported in this study are the same transporter is unknown.

The molecular mechanism by which HPIs inhibit BCRP transport is unknown. However, the finding that a molecule that binds to and inhibits a drug transporter but itself is not transported is not unexpected. Zhang et al. (2000) found that saquinavir inhibits transport of tetraethylammonium by the human organic cation transporter OCT1, but itself is not transported by OCT1. Similarly, the nucleoside analog chloroadenosine was reported to inhibit transport of nucleoside substrates by the human intestinal hCNT2 and hCNT1 Na⁺-nucleoside transporters, but itself is not a substrate of these transporters (Lum et al., 2000; Patil et al., 2000). Most recently, Nakanishi et al. (2003a) studied reciprocal inhibition of BCRP transport by its substrates, including daunorubicin, mitoxantrone, and flavopiridol and found that, in any pair of two substrates, none of the substrate caused mutual inhibition of the transport of the other. For example, flavopiridol significantly inhibited BCRP (482T)-mediated transport of daunorubicin, but daunorubicin did not reciprocally inhibit BCRP (482T) transport of flavopiridol. This led them to propose that BCRP may contain multiple ligand interaction sites. Thus, it is possible that MX and the HPIs have distinct or overlapping binding sites on BCRP.

The finding that ritonavir, saquinavir, and nelfinavir are effective inhibitors but not substrates of BCRP is clinically significant with respect to drug-drug interactions with HPIs. Although all the HPIs are potent inhibitors of CYP3A enzymes and their inhibitory effectiveness to these enzymes is similar, they produce profoundly different degrees of clinically significant drug interactions (Unadkat and Wang, 2000). One possible explanation of these differences is the possibility that the HPIs inhibit multiple and different transporters in the intestine and liver with different potencies. Because BCRP is highly expressed in the intestine, one such transporter could be BCRP. The current recommended doses of ritonavir, saquinavir, and nelfinavir for monotherapy are, respectively, 600 mg twice daily, 600 mg three times daily, and 750 mg three times daily. Thus, assuming 100% dissolution of these drugs in the intestine, the estimated intestinal luminal concentrations of these HPIs on consuming these doses would be around 500 to 1500 μM. Such concentrations far exceed the IC₅₀ values of the HPIs for BCRP. Thus, given the IC₅₀ values observed here (12–20 μM for wild-type BCRP), inhibition of intestinal BCRP activity by the orally administered HPIs could be achieved. Because HPIs extensively bind to protein in plasma, the IC₅₀ values reported in this study may be overestimated as they were determined in the presence of FBS. However, it is unlikely that the blood/plasma concentrations of HPIs will be sufficient to inhibit BCRP systemically because the average steady-state plasma concentrations of the HPIs are approximately 1 to 5 μM (Unadkat and Wang, 2000). Further studies are needed to determine whether drugs routinely administered to HIV patients are substrates of BCRP.

In summary, the present study demonstrates that ritonavir, saquinavir, and nelfinavir are clinically significant inhibitors of BCRP, but indinavir and amprenavir are not. None of the HPIs investigated in this study is a substrate for BCRP. Further studies are needed to determine whether the HPIs are also able to inhibit BCRP-mediated transport of a model substrate drug in in vivo studies using animal models. Such studies are in progress in our laboratory. Inhibition of BCRP transport by HPIs may prove beneficial in increase of the bioavailability of poorly bioavailable drugs that are BCRP substrates.